From 6caf2b8c68efbdced21a45b8aef8220ac2e56fb3 Mon Sep 17 00:00:00 2001
From: Nicolas Boisselier <nicolas.boisselier@gmail.com>
Date: Mon, 23 Sep 2024 10:47:47 +0200
Subject: [PATCH] --nohref

---
 bin/html2csv | 133 ++++++++++++++++++++++++++++-----------------------
 1 file changed, 73 insertions(+), 60 deletions(-)

diff --git a/bin/html2csv b/bin/html2csv
index 8e190bcc..bf0e6a94 100755
--- a/bin/html2csv
+++ b/bin/html2csv
@@ -28,14 +28,14 @@ my $VERBOSE = $main::VERBOSE = 1;
 my $DEBUG = $main::DEBUG = 0;
 
 my %Opt = (
-  'num' => undef,
-  'print-num' => 0,
-  'tag' => 'table',
-  'exp1'=> '',
-  'exp2'=> '',
-  'exp3'=> '',
-  'sep' => "\t",
-  'html' => 0,
+	'num' => undef,
+	'print-num' => 0,
+	'tag' => 'table',
+	'exp1'=> '',
+	'exp2'=> '',
+	'exp3'=> '',
+	'sep' => "\t",
+	'html' => 0,
 );
 get_options(\%Opt);
 # NB 15.05.23 help() unless @ARGV;
@@ -53,17 +53,17 @@ $_ = $Opt{sep}; $html =~ s/$_/ /g;
 $html =~ s/\s+/ /g;
 
 my %TAGS = (
-  'table' => [
-    'table',
-    'tr',
-    't[dh]',
-  ],
-  'dl' => [
-    'dl',
-    '',
-    'd[td]',
-    2,
-  ],
+	'table' => [
+		'table',
+		'tr',
+		't[dh]',
+	],
+	'dl' => [
+		'dl',
+		'',
+		'd[td]',
+		2,
+	],
 );
 exists $TAGS{$Opt{tag}} or die "$NAME: Unknown tag '$Opt{tag}'\n";
 my ($T1,$T2,$T3,$COUNT) = @{ $TAGS{$Opt{tag}} };
@@ -76,47 +76,59 @@ my $exp3 = $Opt{exp3}; $exp3 and $exp3 = "[^>]*$exp3";
 
 for my $table ($html =~ m,<${T1}${exp1}[^>]*>(.*?)<\s*/\s*${T1}\s*>,gi)
 {
-  $table_num++;
-  #warn $table_num;
-  next if defined $Opt{'num'} and $Opt{num} and $Opt{num} != $table_num;
-
-  $table = "<>$table</>" unless $T2;
-  for my $tr ($table =~ m,<(${T2})${exp2}[^>]*>(.*?)<\s*/\s*${T2}\s*>,gi)
-  {
-    my @col;
-    print "$table_num " if $Opt{'print-num'};
-
-  	$tr = "<>$tr</>" unless $T3;
-  	my $count = 0;
-
-    #@_ = $tr =~ m,(<td[^>]*>)\s*<a\s+href="([^"]+)"[^>]*>(.*?)</a>,;
-    #die @_ if @_;
-    $tr =~ s,<td[^>]*>\s*<a\s+[^>]*href="([^"]+)"[^>]*>(.*?)</a>,<td>$1</td><td>$2,g
-    	if $Opt{href2td};
-
-    for my $td ($tr =~ m,<${T3}${exp3}[^>]*>(.*?)<\s*/\s*${T3}\s*>,gi)
-    {
-    $td =~ s,<a\s+[^>]*href="([^"]+)"[^>]*>(.*?)</a>,[href=$1]$2,g
-    	if $Opt{href2txt};
-      $td = html2txt($td) unless $Opt{html};
-      $_ = chr(194).chr(160); $td =~ s/$_/ /g;
-      $td =~ s/\s+/ /g;
-      $td = str_trim($td);
-      $count++;
-
-      push(@col,$td);
-      if ($COUNT and $count > $COUNT)
-      {
-    		print join($Opt{sep},@col)."\n";
-    		@col = ();
-      }
-    }
-
-    print join($Opt{sep},@col)."\n" if @col;
-    #print "\n" unless $T3;
-
-  }
-  #print "\n" unless $T2;
+	$table_num++;
+	#warn $table_num;
+	next if defined $Opt{'num'} and $Opt{num} and $Opt{num} != $table_num;
+
+	$table = "<>$table</>" unless $T2;
+	for my $tr ($table =~ m,<(${T2})${exp2}[^>]*>(.*?)<\s*/\s*${T2}\s*>,gi)
+	{
+		my @col;
+		print "$table_num " if $Opt{'print-num'};
+
+		$tr = "<>$tr</>" unless $T3;
+		my $count = 0;
+
+		#@_ = $tr =~ m,(<td[^>]*>)\s*<a\s+href="([^"]+)"[^>]*>(.*?)</a>,;
+		#die @_ if @_;
+		$tr =~
+			s,<td[^>]*>\s*<a\s+[^>]*href="([^"]+)"[^>]*>(.*?)</a>,<td>$1</td><td>$2,g
+		if $Opt{href2td};
+
+		for my $td ($tr =~ m,<${T3}${exp3}[^>]*>(.*?)<\s*/\s*${T3}\s*>,gi)
+		{
+			$td =~ s,<a\s+[^>]*href="([^"]+)"[^>]*>(.*?)</a>,[href=$1]$2,g
+				if $Opt{href2txt};
+
+			my $hrefTxt = '';
+			if ($Opt{nohref})
+			{
+				$td =~ s,<a\s+[^>]*href="([^"]+)"[^>]*>(.*?)</a>,,g;
+				$hrefTxt = $2;
+				#$td = $2 if $td =~ /^\s*$/;
+			}
+
+			$td = html2txt($td) unless $Opt{html};
+			$_ = chr(194).chr(160); $td =~ s/$_/ /g;
+			$td =~ s/\s+/ /g;
+			$td = str_trim($td);
+
+			$td = $hrefTxt if $Opt{nohref} and $td eq '' and $hrefTxt;
+			$count++;
+
+			push(@col,$td);
+			if ($COUNT and $count > $COUNT)
+			{
+				print join($Opt{sep},@col)."\n";
+				@col = ();
+			}
+		}
+
+		print join($Opt{sep},@col)."\n" if @col;
+		#print "\n" unless $T3;
+
+	}
+	#print "\n" unless $T2;
 
 }
 
@@ -233,6 +245,7 @@ Quick usage:
  -option[tag|T=s] table|dl Default: table
  -option[href2td!] Treat first column href as a new column
  -option[href2txt!] 
+ -option[nohref!] Don't add href to column
  -option[sep|s=s] Default: tab (\t)
  -option[html!] Keep html entities (default: yes)
  -option[num|n=i] Only dump table number
-- 
2.47.3